erlsna is a proof-of-concept application for my Computer Scince undergraduate thesis: Social Network Analysis software for longitudinal node experience analysis on dynamic networks. paper
erlsna follows Agent-based model where individuals (vertices on graph) are mapped to agent processes.
Though not highlighed in detail in the paper, erlsna provides a PubSub mechanism for real-time notifications on individuals. See below.
erlsna has limited functionality and lacks documentation. Feel free to reach out to me if you need assistance. I intend to improve on this subject whenever I can.
-
Erlang R13B03 or newer (not tested with older releases)
-
Rebar (not mandatory but advised)
-
TCP port 8082 (configurable in erlsna.app.src)
Python3 (for python driver and nntp_collector) MongoDB (for nntp_collector storage) R (for some charts) & zoo package (I think)
./autorebar # rebar clean, compile and generate
./rel/mynode/bin/mynode console
For info on TCP message syntax, see the paper, APPENDIX B. For information on analyses and example graphs, see the paper, Chapter 5.
import os
import erlsna
api = erlsna.ERLSNA(live=True)
png_path = os.path.expanduser("~/engagement_graph.png")
api.engagement_analysis("foo@bar.com", with_graph_png=png_path)
For available keywords and further driver uses, refer to the erlsna.py source code.
This was developed for a demo. Currently PubSub is only available for Closeness Centrality metric. One can subscribe to an agent (through Python driver or raw TCP message over Telnet) with a condition and get notified whenever it holds true. An example:
# Telnet
subscribe abc@example.com > 0.123
Will notify the Telnet client whenever the agent's ("abc@example.com") closeness centrality goes over 0.123. The publish will only happen once per going over threshold.
api = ERLSNA()
api.subscribe("abc@example.com", "<", 0.02)
-
: goes over given value
- < : goes below given value
- = : becomes equal to given value
-
- : every time a change happens (a dummy value is expected after eg:
subscribe abc@example.com * 0
)
- : every time a change happens (a dummy value is expected after eg:
Used to collect data from NNTP servers for testing and demonstration, requires MongoDB to be running.
collect: collect (from where it left off) new articles from given server & group emailfixer: to fix dual-line "From:" header issue in NNTP responses, will prompt user for fixes livedump: push data to running erlsna node